Journal of the Formosan Medical Association最新文献

Background: Accurate biopsy-based detection of Helicobacter pylori (H. pylori) is challenging in atrophic gastritis (AG) and intestinal metaplasia (IM). Histologic severity of AG or IM reduces detection rates, but how endoscopic classifications such as Kimura-Takemoto and Endoscopic Grading of Gastric Intestinal Metaplasia (EGGIM) should guide real-time biopsy site selection is uncertain. We evaluated whether these endoscopic features can optimize biopsy strategies for H. pylori.

Methods: This retrospective study analyzed patients with histologically confirmed H. pylori undergoing gastroscopy between June 2023 and September 2024 at a tertiary center in Taiwan. Kimura-Takemoto classification and EGGIM scores (1-4: focal IM; 5-10: extensive IM) were assessed using high-resolution white-light and narrow-band imaging. Biopsies from antrum and corpus were analyzed to determine site-specific detection rate stratified by endoscopic severity.

Results: Among 127 patients (63% female; mean age 61.6 years), the overall detection rate was higher in the antrum than the corpus (89.8% vs. 59.1%, p < 0.001) and remained superior across all stages of Kimura-Takemoto classification. However, with increasing IM severity by EGGIM, antral detection declined while corpus detection increased (focal IM: 92.7% vs. 54.5%, p < 0.001; extensive IM: 84.4% vs. 68.8%, p = 0.302). Additionally, patients aged ≥65 years showed significantly reduced antral detection compared to younger patients (80.8% vs. 96.0%, p = 0.013), whereas corpus detection remained unaffected by age.

Conclusions: Endoscopic features can guide biopsy site selection. Although antral biopsies provide the highest yield, adding corpus biopsies is essential for patients with extensive IM (EGGIM ≥5) or age ≥65 years to optimize H. pylori detection.

Background: Esophageal squamous cell carcinoma (ESCC) poses a significant health burden globally, particularly in Asian regions, with limited therapeutic options and a challenging prognosis. Neoadjuvant chemoradiotherapy (CRT) has emerged as a primary strategy for managing locally advanced ESCC, although the optimal chemotherapy regimen remains unclear. This study aims to compare the efficacy and safety of two commonly used neoadjuvant chemotherapy regimens, 5-fluorouracil plus cisplatin and paclitaxel plus cisplatin, in patients with locally advanced ESCC.

Methods: This retrospective study included ESCC patients who underwent neoadjuvant CRT, either with 5-fluorouracil/cisplatin or paclitaxel/cisplatin, followed by surgery at a single institution in Taiwan between 2019 and 2022. Treatment-related outcomes, pathological responses, survival analyses, and safety profiles were evaluated and compared between the two chemotherapy groups.

Results: Seventy-two patients were analyzed, with 51 (70.8 %) receiving 5-fluorouracil/cisplatin and 21 (29.2 %) receiving paclitaxel/cisplatin. Both groups showed similar rates of downstaging and pathological complete response (pCR). One-year survival rates were 80.4 % for 5-fluorouracil/cisplatin and 90.5 % for paclitaxel/cisplatin. Two-year survival rates were 66.5 % and 60.8 %, respectively. The hazard ratio for overall survival was 0.81 (95 % CI, 0.36-1.79; p = 0.592). No significant differences in toxicity were observed between the groups.

Conclusion: The study findings suggest that both 5-fluorouracil/cisplatin and paclitaxel/cisplatin regimens demonstrate comparable efficacy and safety profiles as neoadjuvant chemotherapy options for locally advanced ESCC. Decisions regarding the choice of regimen should consider factors such as treatment accessibility, financial costs, and patient preferences. Further research, including randomized controlled trials, is warranted to validate these findings and guide clinical decision-making effectively.

Background: Early identification of individuals at elevated cardiovascular risk using routine health examination data is essential for preventive cardiology. Machine learning (ML) offers a scalable and non-invasive approach to enhance risk stratification.

Methods: This retrospective study analyzed 899 asymptomatic adults (mean age: 57.3 ± 11.8 years; 608 men) who underwent coronary artery calcium (CAC) scanning or coronary computed tomography angiography (CCTA) at a health management center in Taiwan between 2018 and 2021. Participants were classified into four CAC-based risk categories (0, 1-99, 100-299, ≥300). Nineteen demographic and clinical variables were used to train decision tree (DT), random forest (RF), and support vector machine (SVM) classifiers. Model performance was evaluated using accuracy and AUC, with AUC differences assessed by DeLong's test.

Results: The RF model demonstrated the best performance (accuracy: 76 %, AUC: 0.78), followed by SVM (accuracy: 70 %, AUC: 0.78) and DT (accuracy: 74 %, AUC: 0.75). All models showed clinically meaningful discrimination using readily accessible, non-laboratory health examination data.

Conclusion: ML models incorporating routine health examination variables can effectively predict CAC-defined cardiovascular risk and may serve as practical, scalable pre-screening tools within preventive healthcare workflows, particularly in settings where laboratory testing or advanced imaging resources may be limited.